2012] 1211

SOCIAL NETWORKS, ADVERTISING, AND ANTITRUST

Catherine Tucker and Alexander Marthews

INTRODUCTION

In the last five years, social media has become a dominant force on

the Internet. Social networking websites, the most popular of which is Fa-

cebook, accounted for 20 percent of all time spent on the Internet in 2010.1

However, it is not clear that so far this extraordinary growth has cre-

ated an antitrust issue. There are two reasons for this. First, consumers do

not pay for using these services on most social networking sites.2 Second,

firms are not spending large amounts of money trying to access consumers

using these new channels.3 This is despite the fact that there are obvious

marketing applications for a form of media that allows companies to ob-

serve potential consumers’ social interactions.4

The reason that firms have not spent much money is not because they

think that social media is not important. Instead, it is because most firms’

commercial strategies for social media have emphasized the success of

“earned reach,” a strategy by which a brand develops its pool of subscribers

organically and also aims to expand that pool by seeing existing subscribers

share links with their social networks.5 Recent academic research, however,

Associate Professor of Marketing at MIT Sloan School of Management, Cambridge, MA and

Faculty Research Fellow at the National Bureau of Economic Research (“NBER”). This paper refers to

academic research that I have conducted that has been supported by Google, NBER, the National Sci-

ence Foundation, the Net Institute, and WPP. I am acting as an expert witness for Facebook on an unre-

lated matter. All errors are my own.

Executive Director of the Economics of Data Privacy Research Institute; M.P.P., University of

California, Berkeley.

1 Facebook accounted for one out of every seven minutes spent online as well as three out of

every four minutes spent on social networking websites. It’s a Social World: Top 10 Need-to-Knows

About Social Networking and Where It’s Headed, COMSCORE, 4, 8 (Dec. 21, 2011), http://www.

comscore.com/Press_Events/Presentations_Whitepapers/2011/it_is_a_social_world_top_10_need-to-

knows_about_social_networking (presentation available for download).

2 E.g., FACEBOOK, http://www.facebook.com (last visited July 7, 2012) (noting registration is

“free and always will be”).

3 See April 2009: Worldwide Social Network Ad Spending, INTERACTIVE ADVER. BUREAU,

http://www.iab.net/insights_research/industry_data_and_landscape/1675/804264 (last visited July 7,

2012); PricewaterhouseCoopers LLP, IAB Internet Advertising Revenue Report, INTERACTIVE ADVER.

BUREAU 3 (Apr. 2010), http://www.iab.net/media/file/IAB-Ad-Revenue-Full-Year-2009.pdf.

4 Peter Wylie, 5 Ways to Use Social Data to Grow Your Business, SOC. MEDIA EXAM’R (Jan. 27,

2011), http://www.socialmediaexaminer.com/5-ways-to-use-social-data-to-grow-your-business.

5 Catherine Tucker, Social Advertising 2 (Feb. 15, 2012) (unpublished manuscript), available at

http://ssrn.com/abstract=1975897; see also Sean Corcoran, Defining Earned, Owned and Paid Media,

1212 GEO. MASON L. REV. [VOL. 19:5

has called into question this non-commercial strategy. Eytan Bakshy et al.

emphasize that the kind of organic sharing that is presupposed by a non-

commercial strategy is far more uncommon than previously supposed, and

that only rarely are commercial messages reliably conveyed via social net-

works.6 A recent study shows that to attain the goal of virality, an advertiser

may be forced to forfeit its message’s commercial usefulness.7 Similarly,

another study shows that when firms try to establish an organic social me-

dia presence, they are more often successful at engaging their own employ-

ees rather than their client base.8

Consequently, advertisers may be required to communicate their

commercial message on social networks through the medium of paid adver-

tising.9 A new type of advertising, called “social advertising,” has recently

appeared on sites like Facebook and LinkedIn.10 “A social ad is an online ad

that ‘incorporates user interactions that the consumer has agreed to display

and be shared. The resulting ad displays these interactions along with the

user’s persona (picture and/or name) within the ad content.’”11 The devel-

opment of the social ad represents a profound technological advancement

for advertisers.12 Advertisers now have the ability to harness the force of an

individual’s social network to target advertising and attract their audience.13

Other work has shown that such “social advertising” can double the

effectiveness of an advertiser’s ad.14 This increase in efficacy bears signifi-

cance because, in recent years, advertisers have often dismissed social net-

working websites as simply sites hosting “paid media”—in other words,

paid advertising. 15 The popular and marketing press have contributed to

this with headlines like “Online Social Networks and Advertising Don’t

FORRESTER (Dec. 16, 2009), http://blogs.forrester.com/interactive_marketing/2009/12/defining-earned-

owned-and-paid-media.html.

6 Tucker, supra note 5, at 2; accord Eytan Bakshy et al., Everyone’s an Influencer: Quantifying

Influence on Twitter, MICH. INTERACTIVE & SOC. COMPUTING (Dec. 10, 2010), http://misc.si.umich.edu

/media/papers/wsdm333w-bakshy.pdf.

7 Tucker, supra note 5, at 2; accord Catherine Tucker, Ad Virality and Ad Persuasiveness 31

(NET Inst., Working Paper No. 11-06, 2011), available at http://ssrn.com/abstract=1952746.

8 Amalia R. Miller & Catherine Tucker, Active Social Media Management: The Case of Health

Care 2 (Jan. 13, 2012) (unpublished manuscript), available at http://ssrn.com/abstract=1984973.

9 Tucker, supra note 5, at 2.

10 Id.; see also Riva Richmond, LinkedIn’s Social-Ad Misstep, N.Y. TIMES (Aug. 17, 2011, 11:31

AM), http://gadgetwise.blogs.nytimes.com/2011/08/17/linkedins-social-ad-misstep.

11 Tucker, supra note 5, at 2 (quoting IAB Social Advertising Best Practices, INTERACTIVE

ADVER. BUREAU 4 (May 2009), available at http://www.iab.net/media/file/Social-Advertising-Best-

Practices-0509.pdf).

12 Id.

13 Id.

14 Id. at 12.

15 Id. at 24.

2012] SOCIAL NETWORKS, ADVERTISING, AND ANTITRUST 1213

Mix” and “Facebook Ad Click-Through Rates Are Really Pitiful.”16 The

results of one study suggest, however, that “as social advertising develops

this will change swiftly.”17 Specifically, social networks will have the abil-

ity to use their access to unique social network data to enlarge their share

of advertising dollars.18

This Article asks whether a rapid expansion of social advertising will

have potential antitrust implications. Part I of this Article presents defini-

tions of social network websites and of the relevant market definition for

advertising on social networking websites. Part II discusses network effects

in social media advertising. Finally, Part III addresses the issue of antitrust

and privacy in social networks.

II. WHAT IS THE MARKET?

A. Defining a Social Media Website

According to comScore,

Social networking accounted for nearly 1 in every 5 minutes spent online globally in October 2011, ranking as the most engaging online activity worldwide. Social networking sites now

reach 82 percent of the world’s Internet population age 15 and older that accessed the Inter-

net from a home or work computer, representing 1.2 billion users around the globe.19

Statistics like these suggest an impressive reach. They also suggest

that it is easy to classify what is and what isn’t a “social media website.”

However, on reflection, the definition is not so clear-cut—as discussed by

Professor Spencer Waller, social networking sites tend to have an “I’ll

know it when I see it” flavor.20 Straightforward definitions that come to

mind, such as “Internet websites that allow users to interact socially,” are

problematic because they are too broad. Many websites, such as Yelp,

16 Id.; Darren Barefoot & Julie Szabo, Facebook Ad Click-Through Rates Are Really Pitiful,

FRIENDS WITH BENEFITS (Apr. 7, 2008, 5:03 PM), http://www.friendswithbenefitsbook.com/

2008/04/07/facebook-ad-click-through-rates-are-really-pitiful; Mitch Joel, Online Social Networks and

Advertising Don’t Mix, SIX PIXELS OF SEPARATION (Feb. 8, 2008, 8:33 PM), http://www.twistimage

.com/blog/archives/online-social-networks-and-advertising-dont-mix.

17 Tucker, supra note 5, at 24.

18 Id.

19 Press Release, comScore, Inc., It’s a Social World: Social Networking Leads as Top Online

Activity Globally, Accounting for 1 in Every 5 Online Minutes (Dec. 21, 2011), available at

http://www.comscore.com/Press_Events/Press_Releases/2011/12/Social_Networking_Leads_as_Top_O

nline_Activity_Globally.

20 See Spencer Weber Waller, Antitrust and Social Networking, 91 N.C. L. REV. (forthcoming

2012) (manuscript at 6-7), available at http://ssrn.com/abstract=1948690.

1214 GEO. MASON L. REV. [VOL. 19:5

allow users to interact socially but would not typically count as “social

media websites.”21

Therefore, this Article uses a very loose definition that says that so-

cial network websites are ones that primarily exist to allow users to post a

profile and exchange or broadcast messages and information with their

friends and contacts. This would include websites like Facebook, Pinterest,

Google+, Orkut, LinkedIn, Twitter, and Myspace.22 External to this defini-

tion would be websites such as YouTube or Yelp that have a social ele-

ment but have another primary form of user-generated content (be it videos

or reviews) that drives users to the website.

Even within this definition, it should be noted that there are very

different types of social networking sites. A key distinction is that some

social networking sites, like Facebook, focus on sharing and symmetric

exchange between users, and some, like Twitter, promote asymmetric ex-

change between users and consequently follow more of a broadcast mod-

el.23 There are also sites where users tend to post information passively,

such as LinkedIn, compared to sites that require more daily engagement,

like Pinterest.24

B. Market Definition

The question of what is a social networking website is not quite the

same question as what is the relevant market definition for advertising on

social networking websites. Market definition is important when determin-

ing whether a particular social networking website has a monopoly over

advertising.25 There may be many other forms of advertising that are part of

the same market, even if they are not explicitly ads on social networks.

The first question is whether advertising markets that are not specifi-

cally online should be considered as part of the relevant market. For a more

detailed discussion of the issues, a recent paper specifically discusses the

extent to which online advertising can be considered as a market that is

21 Id. (manuscript at 7).

22 Formerly known as “MySpace,” the company modified its name to “Myspace” in 2010. Press

Release, Myspace, Inc., Meet the New Myspace (Oct. 27, 2010), available at http://www.myspace.com/

pressroom/2010/10/meet-the-new-myspace. This Article refers to the company as Myspace, unless an

article title, case caption, or quote uses the old name.

23 Joshua Porter, Relationship Symmetry in Social Networks: Why Facebook Will Go Fully Asym-

metric, BOKARDO (Mar. 29, 2009), http://bokardo.com/p/1006.

24 Alex Wilhelm, Getting Started with LinkedIn, THE NEXT WEB (May 22, 2011),

http://thenextweb.com/insider/2011/05/22/getting-started-with-linkedin; see Mark W. Smith, How to

Use Pinterest’s Pinboard for the Web, USA TODAY, http://www.usatoday.com/tech/news/story/2012-

01-17/how-to-pinterest-mark-smith/52615856/1 (last updated Jan. 17, 2012, 3:06 PM).

25 See Waller, supra note 20 (manuscript at 5).

2012] SOCIAL NETWORKS, ADVERTISING, AND ANTITRUST 1215

isolated from offline advertising markets.26 The paper discusses two studies,

which establish substitution patterns between offline and online markets

both in terms of price and advertising effectiveness.27 In both cases, the

authors present evidence that offline advertising markets do regulate the

operation of online advertising markets.28 The evidence is mainly sugges-

tive, but at the very least it supports the idea that online advertising markets

should not be considered in isolation.

The second question is whether (and which) other online advertising

markets should be part of the relevant market definition. Very loosely, the

online advertising market is divided between search advertising and display

advertising.29 Search advertising—that is, ads that run alongside a search

engine query—accounts for half of all online ad spending, and display

(banner) advertising accounts for the other half.30 Usually, ads on social

networks are counted as part of display advertising.31 Table 1 summarizes

the market share of display ad impressions of different websites. Obviously,

social media is not close to being dominant in display advertising even if it

is the largest server of ad impressions. It is also not clear why search adver-

tising would not be an obvious competitor for display advertising in social

media. In both cases, search engines and social networks use expansive data

on their users’ actions to serve relevant and timely ads.32

26 Avi Goldfarb & Catherine Tucker, Substitution Between Offline and Online Advertising Mar-

kets, 7 J. COMPETITION L. & ECON. 37, 38 (2011).

27 Id. at 40-41; Avi Goldfarb & Catherine Tucker, Advertising Bans and the Substitutability of

Online and Offline Advertising, 48 J. MARKETING RES. 207, 208 (2011); Avi Goldfarb & Catherine

Tucker, Search Engine Advertising: Channel Substitution When Pricing Ads to Context, 57 MGMT. SCI.

458, 459-60 (2011).

28 Goldfarb & Tucker, supra note 26, at 38.

29 Federico Etro, Understanding Online Advertising: Market Structure and Leadership, EUR. BUS.

REV., Mar.-Apr. 2012, at 96, 96, available at http://www.intertic.org/Policy%20Papers/EBR2.pdf.

30 See Tiernan Ray, GOOG: Nomura Starts at Buy; Aspects of a Telco Company, BARRON’S

(Mar. 5, 2012, 5:20 PM), http://blogs.barrons.com/techtraderdaily/2012/03/05/goog-nomura-starts-at-

buy-really-a-telco-company.

31 See, e.g., Press Release, comScore, Inc., U.S. Online Display Advertising Market Delivers 1.1

Trillion Impressions in Q1 2011 (May 4, 2011), available at http://www.comscore.com/Press_Events/

Press_Releases/2011/5/U.S._Online_Display_Advertising_Market_Delivers_1.1_Trillion_Impressions_

in_Q1_2011.

32 Wylie, supra note 4.

1216 GEO. MASON L. REV. [VOL. 19:5

Table 1: Top 10 U.S. Online Display Ad Publishers (Q1 2011)33

Ad Impressions

(billions)

% Share

Facebook 346,455 31.2%

Yahoo! Sites 112,511 10.1%

Microsoft Sites 53,592 4.8%

AOL, Inc. 33,454 3.0%

Google Sites 27,993 2.5%

Turner Digital 18,050 1.6%

Fox Interactive Media 11,697 1.1%

Glam Media 10,207 0.9%

CBS Interactive 9,208 0.8%

Viacom Digital 9,051 0.8%

Evidence for the substitutability of search engine advertising, display

advertising, and advertising on social networks is mainly anecdotal but rea-

sonably compelling. Indeed, there are companies devoted to facilitating the

easy transfer of advertising dollars between these three media. For exam-

ple, HubSpot helps advertisers optimize between search, display and social

network advertising by providing comparative return on investment figures

between all three advertising types to help advertisers reallocate dollars

towards the most effective medium for their product.34 Figure 1 provides a

screenshot of a typical user interface screen at HubSpot. The extent to

which the digitization of data and advertising metrics has led to easy com-

parisons and substitution across media is evident. And the ease of substitu-

tion has important implications for likely market definition.35

Therefore, in general, though this Article uses a relatively narrow se-

mantic definition of what a social networking site is for the purposes of

discussion, at least for advertising markets, the true market definition for

antitrust purposes seems likely to be far broader.

33 Press Release, comScore, Inc., supra note 31.

34 HubSpot Fact Sheet, HUBSPOT, http://www.hubspot.com/Portals/53/docs/HubSpot%20Fact%

20Sheet%20-%20Print%20Version%20-%20Updated%206-7-11.pdf (last visited July 7, 2012). Disclo-

sure: HubSpot is linked to MIT Sloan School of Management.

35 See Waller, supra note 20 (manuscript at 5, 7-9).

2012] SOCIAL NETWORKS, ADVERTISING, AND ANTITRUST 1217

Figure 1: HubSpot Facilitates the Transfer of Advertising Dol-

lars Based on Return on Investment (ROI)36

II. ANTITRUST ISSUES IN SOCIAL MEDIA ADVERTISING

A. Standard Models of Network Effects in Advertising Markets

To understand the potential antitrust issues that may arise in advertis-

ing markets on social media platforms, it is important to summarize what

standard economic theory says about network effects in such markets.37

Network effects occur, quite simply, when the value of a product increases

through a rise in the number of people using it.38 Very loosely, network

effects can either be direct—that is, there is a direct performance benefit for

users as more users utilize the network, such as is the case with video-

messaging services like Skype—or network effects can be indirect, where

the presence of one group of users (such as software developers) benefit

36 Christopher Haddad, How to Finish Your Monthly Marketing Deck in 10 Minutes, HUBSPOT

BLOG (Mar. 2, 2011, 12:00 PM), http://blog.hubspot.com/blog/tabid/6307/bid/10173/How-to-Finish-

Your-Monthly-Marketing-Deck-in-10-Minutes.aspx.

37 Cf. Steven C. Salop & R. Craig Romaine, Preserving Monopoly: Economic Analysis, Legal

Standards, and Microsoft, 7 GEO. MASON L. REV. 617, 638-39 (1999) (discussing the economics of

network effects in United States v. Microsoft Corp.).

38 DANIEL F. SPULBER & CHRISTOPHER S. YOO, NETWORKS IN TELECOMMUNICATIONS:

ECONOMICS AND LAW 120 (2009).

1218 GEO. MASON L. REV. [VOL. 19:5

another group of users (such as software users).39 Network effects can also

be referred to as “network externalities” if there is evidence that the net-

work owner’s pricing strategy does not reflect these network effects.40

Network effects are potentially of importance in antitrust because they

could possibly impede competition. For example, in United States v. Mi-

crosoft Corp.41 it was argued that there were network effects for an operat-

ing system, as the value of the operating system grew with both its number

of users and the number of firms consequently developing applications for

it.42 It was then argued that this meant that Microsoft did not face competi-

tion in the operating system market, and that it used this incumbency to try

and dominate related markets such as browsers.43 In general, the reason that,

in network industries, increased antitrust scrutiny may be warranted is “the

theory that predation, via predatory pricing, ties, exclusive dealing, or the

closure of access to an essential facility, is much easier to achieve in a net-

work.”44

In works by Professor Mark Armstrong and by Professors Jean-

Charles Rochet and Jean Tirole, advertising markets are modeled as having

network effects that are unusual in two respects.45 First, advertising markets

have negative indirect effects. Typically, two-sided markets (such as game

developers and game users, or bank card users and merchants) are modeled

as having strong positive indirect network effects, where one group values

the presence of the other group.46 By contrast, advertising markets are usu-

ally modeled as having negative indirect effects—that is, readers of content

would weakly prefer in all cases for there to be fewer ads.47 Second, content

readers treat the presence of ads as a nuisance that they accept in order to

access subsidized content.48 There are a few cases where in advertising

markets, network effects are suspected of being positive. These are general-

39 Michael L. Katz & Carl Shapiro, Network Externalities, Competition, and Compatibility, 75

AM. ECON. REV. 424, 424 (1985).

40 STAN J. LIEBOWITZ & STEPHEN E. MARGOLIS, WINNERS, LOSERS & MICROSOFT 68 (1999).

41 84 F. Supp. 2d 9 (D.D.C. 1999).

42 Id. at 20.

43 LIEBOWITZ & MARGOLIS, supra note 40, at 252; Timothy F. Bresnahan, The Economics of the

Microsoft Case 2-3 (Stanford Law Sch. John M. Olin Program in Law & Econ., Working Paper No.

232, 2002), available at http://ssrn.com/abstract_id=304701.

44 Max Schanzenbach, Network Effects and Antitrust Law: Predation, Affirmative Defenses, and

the Case of U.S. v. Microsoft, 2002 STAN. TECH. L. REV. 4, ¶ 3 (2002).

45 Mark Armstrong, Competition in Two-Sided Markets, 37 RAND J. ECON. 668, 681-82, 684

(2006); Jean-Charles Rochet & Jean Tirole, Two-Sided Markets: A Progress Report, 37 RAND J. ECON.

645, 659 (2006).

46 Rochet & Tirole, supra note 45, at 645-46.

47 See Simon P. Anderson & Stephen Coate, Market Provision of Broadcasting: A Welfare Analy-

sis, 72 REV. ECON. STUD. 947, 950 (2005).

48 Id.

2012] SOCIAL NETWORKS, ADVERTISING, AND ANTITRUST 1219

ly occasions when users are explicitly seeking information, such as in the

case of Yellow Pages directories.49

The fact that these network effects are presumed to be negative makes

the antitrust issue in generic advertising markets less clear. To see this, con-

trast the case of a classifieds listing site, where there are presumed to be

positive network effects, and a photo-sharing website that is supported by

advertising, which is presumed to have negative network effects. If the

classifieds website managed to attract all the classifieds ads in the world,

then it might garner some degree of market power over consumers who are

seeking classifieds ads. On the other hand, if the photo-sharing website

attracted all the firms who wanted to advertise, then this would certainly not

give the website a direct advantage in terms of attracting consumers and

might even put it at a disadvantage if consumers left to seek less advertis-

ing-cluttered venues.

Unusually, network effects may still be important in the specific mar-

ket of advertising on social networks because of network effects inherent in

the kind of content shown on social network websites.

B. There Are Definitely Network Effects on the Consumer Side

What is certain about social networks is that there are direct network

effects on the consumer side. When it comes to teaching undergraduates

about direct network effects, economics professors have often used the ex-

ample of a fax machine and pointed out that if someone owns the only fax

machine in the world, they are unlikely to find it useful.50 Social networks

may be an even starker example of direct network effects. If there were

only one person on a social network, then that person would be unlikely to

find it useful or enjoyable.51 Therefore, the value of social networks, as

suggested by their name, undoubtedly increases through a rise in the num-

ber of users.

C. How Much Do Network Effects Matter?

The next question is how likely these network effects are to matter in

terms of increasing a social network’s ability to dominate its competition.

49 Marc Rysman, Competition Between Networks: A Study of the Market for Yellow Pages, 71

REV. ECON. STUD. 483, 483 (2004).

50 E.g., Michael L. Katz & Carl Shapiro, Systems Competition and Network Effects, 8 J. ECON.

PERSP. 93, 96-97 (1994).

51 Indeed, in contrast to a fax machine, they couldn’t even use a social network profile as an

expensive photocopier.

1220 GEO. MASON L. REV. [VOL. 19:5

1. Size

There is little work that has attempted to measure the size of net-

work effects on social networking sites on the consumer side, directly.

However, the literature has produced some suggestive evidence that such

network effects are likely to be large. A study by Professor Michael

Trusov et al. suggests that on a social networking site, the elasticity of sign-

ups with respect to electronic invitations sent to members by current

members is estimated to be 0.53.52 In other words, invitations issued by

current members are very effective at increasing membership.53 However,

again this research only speaks to network effects that have very limited

scope.

The other point that is likely to be true about the size of network ef-

fects in social networks is that they will vary greatly across members and

their connections. A 2008 study shows that the average connection in a

social network has very little influence on the adoption behavior of an indi-

vidual.54 In the firm setting of that study, it was only individuals who were

higher in the firm’s hierarchy (such as managers), and individuals who at-

tained informal influence due to their positions as gatekeepers in the com-

munication structure, who actually affected users’ adoption of the technolo-

gy.55 In a similar manner, one might think that, on a social networking site,

it would be an individual’s close friends and family who ultimately influ-

ence the choice of social networking site, and that other acquaintances may

be far less important.

2. Scope

An important question, which is not asked often enough when consid-

ering whether network effects lead to concerns about competitive intensity

in an industry, is what the scope is of network effects. “Scope” is quite

simply the question of how many people in the network matter for the hap-

piness of its users.56 For example, Facebook has over 900 million users

worldwide, but it is not the case that a user values them all.57 Instead, it is

52 Michael Trusov, Randolph E. Bucklin & Koen Pauwels, Effects of Word-of-Mouth Versus

Traditional Marketing: Findings from an Internet Social Networking Site, 73 J. MARKETING 90, 98

(2009).

53 Id. at 96.

54 See Catherine Tucker, Identifying Formal and Informal Influence in Technology Adoption with

Network Externalities, 54 MGMT. SCI. 2024, 2031 (2008).

55 Id.

56 See MILTON L. MUELLER, JR., UNIVERSAL SERVICE 22-23 (1997) (noting that in telecommuni-

cations markets, scope is more important than scale because of users’ heterogeneous preferences).

57 Statistics, FACEBOOK NEWSROOM (Jan. 23, 2012), http://newsroom.fb.com/Key-Facts/Statistics-

8b.aspx.

2012] SOCIAL NETWORKS, ADVERTISING, AND ANTITRUST 1221

likely that the user values the set of people that she already interacts with,

and the people with whom there is a positive probability that she may inter-

act. If the scope of network effects is small for any one individual user—

even though, following the 2009 study by Professor Trusov et al., they

place relatively high valuations on each one of their contacts’ presence on

the network—then this may consequently reduce the extent to which net-

work effects lead to lock-in for that user and mean that the anticompetitive

implications of network effects are less than supposed.58

3. Persistence

An important question about network effects that often goes unasked

is with respect to their likely persistence. Economic theory posits that in

any market where there are network effects and consumers can exit or enter

easily, market dominance is incredibly unstable.59 In such markets, there is

a continuum of multiple equilibria in which each represents a different level

of adoption.60 Intuitively, imagine a situation where two people can either

coordinate to meet at the Empire State building or at Penn Station in New

York. It is possible to predict that they could choose to adopt different

meeting places at different times, since all that matters for the network ef-

fect is whether the other person is present at that particular point in time.

Table 2 provides some evidence that there are indeed such instances of mul-

tiple “equilibria” in social networking websites across the world.

58 See Trusov, Bucklin & Pauwels, supra note 52, at 93 (discussing the value to network members

of adding more friends to the network). For a general discussion of network effects and lock-in, see Katz

& Shapiro, supra note 50, at 96-99.

59 Cf. United States v. Microsoft Corp., 253 F.3d 34, 55-56 (D.C. Cir. 2001) (en banc) (per curi-

am) (finding Microsoft to have obtained market dominance in part because consumers found it difficult

to switch operating systems).

60 See, e.g., Katz & Shapiro, supra note 50, at 97.

1222 GEO. MASON L. REV. [VOL. 19:5

Table 2: Top Social Network in Individual Asia Pacific Markets by Percent

Reach of Web Population in Early 201061

Country Top Social Network

in Market

% Reach of Web

Population

Indonesia Facebook 84.9%

Philippines Facebook 84.5%

Malaysia Facebook 77.5%

Singapore Facebook 72.1%

Australia Facebook 69.4%

New Zealand Facebook 63.6%

Hong Kong Facebook 62.6%

Taiwan Wretch 62.5%

South Korea CyWorld 54.2%

India Orkut 46.8%

Japan Mixi 18.9%

Vietnam Facebook 18.4%

This example makes clear that network effects are generally not

enough to lead to persistent anticompetitive effects. Instead, network effects

need to be linked with “switching costs,” which prevent the easy entry and

exit of consumers once they have adopted a network standard.62 In the oper-

ating system example from Microsoft, there were switching costs for users

of Windows because it was costly and difficult for people to move their

files and data out of the Windows format to an alternative format.63 There-

fore, to really understand the likely importance of network effects in this

market for antitrust purposes, it is necessary to try and understand how like-

ly and easy users are to find it to switch social networks or try to use a new

social network.64

There are obvious potential sources of switching costs on social net-

works. For example, if someone switches to a new social network, then she

has to recreate her profile and transfer any photos and personalization. She

also has to learn an entirely new interface. However, there is little evidence

that these drawbacks have been severe enough to stop users from switching

social networks, at least in the past. One only has to remember the rapid rise

61 Press Release, comScore, Inc., Social Networking Habits Vary Considerably Across Asia-

Pacific Markets (Apr. 7, 2010), available at http://www.comscore.com/Press_Events/Press_Releases/

2010/4/Social_Networking_Across_Asia-Pacific_Markets.

62 See Joseph Farrell & Carl Shapiro, Dynamic Competition with Switching Costs, 19 RAND J.

ECON. 123, 123 (1988).

63 United States v. Microsoft Corp., 84 F. Supp. 2d 9, 15 (D.D.C. 1999).

64 See U.S. DEP’T OF JUSTICE & FED. TRADE COMM’N, HORIZONTAL MERGER GUIDELINES § 7.2

(2010).

2012] SOCIAL NETWORKS, ADVERTISING, AND ANTITRUST 1223

and fall of Friendster and Myspace in this area to see evidence of the great

fragility of network effects in earlier social networking websites. Friendster

was founded in 2002 and until 2004 it was the largest social networking

website in the United States.65 Myspace largely replaced Friendster because

Myspace offered better personalization options and the potential to hear

new music.66 Myspace itself was the most visited social networking website

between 2005 and April 2008.67 However, its lack of privacy controls, em-

phasis on music, and lack of new features are conjectured to have led it to

be surpassed by Facebook.68 As the industry evolves, it will be interesting to

observe whether Facebook, which is currently the most visited social net-

working site,69 will continue to dominate or whether it will suffer a similar

decline.

For one to postulate that this time is different and that Facebook has a

kind of market power that Myspace did not have, one has to suppose that

there is some mechanism keeping users on Facebook and increasing switch-

ing costs that did not exist for Myspace. In reality, however, users can de-

lete their Facebook profiles relatively easily, so long as they are willing to

spend two weeks ensuring that they do not accidentally log into their ac-

counts, thus reactivating them.70 Another parallel example of the lack of

switching costs, and the ability of users to maintain profiles potentially at

multiple social networking websites, is the experience of Orkut. Orkut is a

social networking site that is owned and operated by Google.71 It achieved

high penetration in Estonia, Brazil and India.72 Generally, in Brazil it is

reputed to be in decline relative to Facebook,73 but as the statistics in Table

3 indicate, even in a period of decline, Orkut’s membership increased.

65 See Leah Lerner, How Friendster Works, HOWSTUFFWORKS, http://computer.howstuffworks

.com/internet/social-networking/networks/friendster.htm (last visited July 7, 2012).

66 Danah M. Boyd & Nicole B. Ellison, Social Network Sites: Definition, History, and Scholar-

ship, 13 J. COMPUTER-MEDIATED COMM. 210, 216-17 (2008).

67 See John Owens, MySpace in Agency Talks Ahead of Brand Relaunch, PRWEEK (May 17,

2012, 8:00 AM), http://www.prweek.com/uk/news/1132132/myspace-agency-talks-ahead-brand-

relaunch.

68 See Felix Gillette, The Rise and Inglorious Fall of Myspace, BLOOMBERG BUSINESSWEEK, June

27, 2011, at 52, 58.

69 See It’s a Social World, supra note 1, at 8.

70 Justin Phelps, How to Delete Your Facebook Account, PCWORLD (Nov. 3, 2011, 3:00 PM),

http://www.pcworld.com/article/242956/how_to_delete_your_facebook_account.html.

71 Ricardo Geromel, Facebook Surpasses Orkut, Owned by Google, in Numbers of Users in Bra-

zil, FORBES.COM (Sept. 14, 2011, 6:33 AM), http://www.forbes.com/sites/ricardogeromel/2011/09/14/

facebook-surpasses-orkut-owned-by-google-in-numbers-of-users-in-brazil.

72 Kaifu Zhang & Miklos Sarvary, A Competitive Analysis of Web 2.0 Communities: Differentia-

tion with User-Generated Content 19 (June 5, 2011) (unpublished manuscript), available at

http://ssrn.com/abstract=1858252.

73 See Geromel, supra note 71.

1224 GEO. MASON L. REV. [VOL. 19:5

Table 3: Selected Social Networking Sites in Brazil by Total Unique

Visitors (000), December 2011 vs. December 201074

Dec. 2010 Dec. 2011 % Change

Total Internet: Total Audience 45,128 51,759 15%

Social Networking 42,477 47,399 12%

Facebook 12,379 36,098 192%

Orkut 32,671 34,419 5%

Windows Live Profile 11,801 13,301 13%

Twitter 8,939 12,499 40%

Vostu 1,120 4,901 338%

Google Plus N/A 4,300 N/A

Tumblr

LinkedIn

1,319

N/A

4,029

3,182

206%

N/A

D. Consumer-Side Network Effects May Lead to Network Effects in

Advertising

The immediate question for competition authorities is whether there

are markets that are “tied” to the consumer social network market, meaning

that social networking websites can benefit from the market’s inherent net-

work effects.75 The most obvious such market is online advertising.76

What is unique about social networks is that the existence of a cus-

tomer network can itself actually be used to improve advertising. One of the

authors has previously investigated a new form of advertising called “social

advertising.”77 In social advertising, marketers use online social relation-

ships to target and improve their ads.78

Previous work by one of the authors has examined data from a field

experiment in which a nonprofit used both traditional and social advertising

on Facebook.79 Data analysis suggests that the social ads, which were tar-

geted to friends of “fans” of the nonprofit on Facebook, attracted far more

74 Press Release, comScore, Inc., Facebook Blasts into Top Position in Brazilian Social Network-

ing Market Following Year of Tremendous Growth (Jan. 17, 2012), available at

http://www.comscore.com/Press_Events/Press_Releases/2012/1/Facebook_Blasts_into_Top_Position_i

n_Brazilian_Social_Networking_Market.

75 See David S. Evans, Antitrust Issues Raised by the Emerging Global Internet Economy, 102

NW. U. L. REV. COLLOQUY 285, 294 (2008), available at http://www.law.northwestern.edu/

lawreview/colloquy/2008/13/LRColl2008n13Evans.pdf.

76 This Article does not consider other markets such as social gaming and “applications.”

77 Tucker, supra note 5, at 1-2.

78 Id. at 2.

79 Id. at 6-7.

2012] SOCIAL NETWORKS, ADVERTISING, AND ANTITRUST 1225

clicks than those that were not.80 When advertisers target ads based on who

is friends with whom, they can double the number of clicks, because adver-

tisers can uncover consumers who may also be interested in their product.81

In sociology, the phenomenon that friends may have similar (unobserved)

tastes is called “homophily” and is sometimes summarized by the expres-

sion “birds of a feather flock together.”82 The use of social network data to

improve advertising relevance and consequently revenue means that net-

work effects may be more important than in other advertising networks.

The obvious question is whether this form of social advertising, be-

cause it uses social network data, is also subject to network effects. This

answer is less clear-cut. Certainly, the value to the advertiser for an ad may

increase with the number of a user’s social connections. This would particu-

larly be the case if the advertiser’s aim were predominantly to obtain a large

reach—that is, a large number of eyeballs for the ad. However, it seems

most likely that the value of such ads to advertisers would increase mainly

in the quality of the network data that the social network offered.83 In other

words, social networks with data that most accurately reflected meaningful

social relationships would be most valuable.

III. ANTITRUST AND PRIVACY IN SOCIAL NETWORKS

It is obligatory when discussing social networks to mention privacy.

This is because social networks have enabled the sharing of digital data at

unprecedented levels, raising natural questions of consumer privacy.84

Privacy has just recently entered antitrust discussions.85 The current

legal debate has focused on whether concerns about privacy should be used

as a criterion to reject proposed mergers.86 In other words, the question is

whether a reduction in consumer surplus due to privacy erosion should be

treated in a similar manner to a reduction in consumer surplus due to higher

80 Id. at 3.

81 See id. at 12, 14.

82 Miller McPherson, Lynn Smith-Lovin & James M. Cook, Birds of a Feather: Homophily in

Social Networks, 27 ANN. REV. SOC. 415, 416-17 (2001).

83 See Mauro Bampo et al., The Effects of the Social Structure of Digital Networks on Viral Mar-

keting Performance, 19 INFO. SYS. RES. 273, 287 (2008).

84 Eli Edwards, Stepping Up to the Plate: The Google-DoubleClick Merger and the Role of the

Federal Trade Commission in Protecting Online Data Privacy 30-31 (Apr. 25, 2008) (unpublished

manuscript), available at http://ssrn.com/abstract=1370734.

85 See Senator Al Franken, Remarks to the American Bar Association (Antitrust Section) 18-27

(Mar. 29, 2012), available at http://assets.sbnation.com/assets/1033745/franken_aba_antitrust_

speech.pdf.

86 See Edwards, supra note 84, at 2.

1226 GEO. MASON L. REV. [VOL. 19:5

prices.87 However, more recent work has questioned whether this is the

right antitrust question relating to privacy. Specifically, Professor James

Campbell et al. ask whether the introduction of privacy regulation may ul-

timately lead to less competition.88 Campbell et al. use an economic theory

model to illustrate this point, but the intuition is relatively straightforward,

as shown by the following thought experiment.89

Imagine it is the late 1990s and you had the choice of the two search

engines shown in Figure 2. AltaVista is the incumbent that you are com-

fortable and familiar with. Google is a recent upstart that has a better prod-

uct but that you know little about.90 Imagine now that privacy regulation

forced users to choose whether to give explicit opt-in consent to these two

search engines upon first use. It is likely that while users might be more

comfortable with giving this consent to the well-known AltaVista, they

would feel less comfortable with giving consent to the upstart Google. In

this manner, privacy regulation can inadvertently favor incumbents and

hamper the operations of start-ups.

87 Cf. id. at 3-4 (noting that the Federal Trade Commission considers privacy violations without a

corresponding benefit to “be unfair practices” in the merger context).

88 James Campbell, Avi Goldfarb & Catherine Tucker, Privacy Regulation and Market Structure 2

(Dec. 9, 2011) (unpublished manuscript), available at http://ssrn.com/abstract=1729405.

89 See id. at 2-5.

90 AltaVista was the incumbent search engine in the mid-1990s, and “the Google of its era.” JOHN

BATTELLE, THE SEARCH: HOW GOOGLE AND ITS RIVALS REWROTE THE RULES OF BUSINESS AND

TRANSFORMED OUR CULTURE 45 (2005).

2012] SOCIAL NETWORKS, ADVERTISING, AND ANTITRUST 1227

Figure 2: Which Would You Choose?

CONCLUSION

People spend 20 percent of their time online on social networking

sites.91 That represents an extraordinary amount of time given that this cate-

gory of website hardly existed seven years ago. However, as of 2012, there

is little indication this has led to explicit antitrust issues. This is partially

because of the difficulty to prove anticompetitive harm when users almost

always experience social networking sites for free. Furthermore, the extent

to which there are antitrust issues in the advertising market, which is the

main place that Facebook has attempted to monetize its user base, appears

for now to be limited, simply because of vigorous competition from other

advertising channels and the inherent fragility of these social networks.

91 It’s a Social World, supra note 1, at 4.